Is MUM negatively affecting Google search rankings? Is it making it easier for Google to answer complex search needs?
Google introduced MUM (Multitask Unified Model) technology at Google I/O 2021 to assist its ranking systems in comprehending language. The question of whether or not MUM will be used as a ranking factor has been up for debate since the announcement.
What Is MUM?
Google claims that MUM is a significant step forward in information processing and understanding. MUM is intended to help Google answer complex queries in search. MUM is 1,000 times more potent than BERT, Google’s previous NLP transfer learning system.
MUM’s T5, Text-To-Text Transfer Transformer model, reframes NLP assignments into a unified text-to-text format to better comprehend information and knowledge.
Google says that MUM could be used for document summarization, question answering, and classification tasks such as sentiment analysis. Given its importance inside the Googleplex, MUM should undoubtedly be on your radar.
MUM As A Ranking Factor
When Google first announced the MUM initiative, many people were concerned about how it might affect search rankings (particularly their own). While most of Google’s ranking algorithms are untouched yearly, some are significant.
According to Google, BERT was the most significant update in five years when it was rolled out globally in 2019. BERT impacted about 10% of search queries, as expected.
Google’s RankBrain algorithm was deployed in spring 2015 to alter SERPs significantly. Since Google recently revealed MUM, SEO professionals and their clients should note it.
In his post, Roger Montti talks about a patent that might provide more information about MUM’s inner workings. It’s an interesting read if you’re interested in what’s happening under the hood. For now, let’s see whether MUM is a ranking factor.
MUM Is Not A Ranking Factor Based On Evidence
Despite MUM technology not being used yet, in May 2021, Google fellow and Search vice president Pandu Nayak made it clear that MUM technology isn’t yet in play:
“Today’s search engines aren’t quite cultivated enough to answer how a professional would. But with a recent technology named Multitask Unified Model, or MUM, we’re getting closer to assisting you with these complex needs. So, you’ll need fewer searches to get things done in the future.”
The timeline for when MUM-enabled features and updates would go live was “in the coming months and years.” Regarding whether the industry would be informed when MUM is available on search, Google Search Liaison Danny Sullivan said ‘yes.’
MUM as A Ranking Factor Based On Evidence
At RankBrain’s launch six months later, it was not announced until after the fact. In general, updates aren’t even announced or confirmed.
Google has improved at sharing impactful updates before they happen, although sharing them has not improved. For instance, BERT was first announced in November 2018, rolled out for English-language queries in October 2019, and subsequently rolled out worldwide in December 2019.
Google announced Page Experience and Core Web Vitals improvements over a year before they were released in June 2021. Google has already indicated that MUM will be a significant development. However, might MUM be linked to lower rankings for countless websites in early spring and late summer 2021?
Improving Search Results by Implementing MUM.
As promised, Google released information about new and prospective MUM applications in June 2021. Vaccine information was improved based on what Google described as the first MUM application in June 2021.
“With MUM, we were able to recognize over 800 variations of vaccine names in more than 50 languages in a matter of seconds. After validating MUM’s results, we used them to Google Search so that people could find timely, high-quality information about COVID-19 vaccines worldwide.”
In February 2022, Google discussed how neural matching, BERT, and MUM assisted information understanding. Google shared in September 2021 how MUM could be used in the future, including visual and textual search – as well as a revamped search page to make it more natural and intuitive. The following is a paraphrase:
“Even though MUM has much to offer, we have already used it to enhance searches for COVID-19 vaccination info. In the coming months, Google Lens will provide more intuitive search methods using text and pictures. These are extremely specialized jobs, so MUM is not employed to boost the quality of search results like RankBrain, neural matching, and BERTs do.”
Google posted an update in March 2022 regarding how MUM applied to searches for personal crises.
“Our newest AI program, MUM, can better detect a wider range of personal emergency searches. MUM can better grasp the purpose behind people’s queries to determine if someone is in trouble, which allows us to provide trustworthy and practical information at the right time. MUM will start to improve our services shortly.”
Google then went on to elaborate on how MUM might enhance search results.
“With MUM, we can scale safety protections worldwide much more efficiently by transferring knowledge across the 75 languages it has been trained in. We train one MUM model to perform one job—for example, classifying queries—and it learns to perform it in all the languages it knows.
For example, we use AI to filter out spam pages in your search results. We will use MUM to improve our spam protection in the coming months and expand to languages with little training data to detect personal crisis queries better worldwide, working with local partners to provide actionable information in more nations.”
Final Thoughts: MUM Could Be Used As A Ranking Factor.
Google has not yet started using MUM as a search ranking signal, but it may in the future. In his posts on The Keyword blog, Nayak insists that MUM will be tested as rigorously as BERT before Google uses it in searches.
Read Next: Google’s Helpful Content Update
For SEO audits, click here to book your appointment.
What’s Your SEO Score?
Enter the URL of any landing page or blog article and see how optimized it is for one keyword or phrase.